A Novel Explainable Fuzzy Clustering Approach for fMRI Dynamic Functional Network Connectivity Analysis.
Charles A Ellis1, Robyn L Miller1, Vince D Calhoun1
1Tri-institutional Center for Translational Research in Neuroimaging and Data Science: Georgia State University, Georgia Institute of Technology, Emory University, Atlanta, GA 30303 USA.
Biorxiv : the Preprint Server for Biology
|February 13, 2023
Summary
This study introduces a novel fuzzy clustering approach for resting-state fMRI analysis, improving the understanding of brain network dynamics in schizophrenia. The method enhances explainability in dynamic functional network connectivity (dFNC) research.
Area of Science:
- Neuroscience
- Computational Biology
- Medical Imaging
Background:
- Resting-state functional magnetic resonance imaging (rs-fMRI) and dynamic functional network connectivity (dFNC) are crucial for studying brain network interactions in neuropsychiatric disorders.
- Current methods using hard clustering to identify brain activity states have limitations in quantifying feature importance across all samples.
Approach:
- A novel approach utilizing fuzzy clustering is proposed to output state probabilities for each sample.
- Kullback-Leibler divergence is employed to measure the impact of feature perturbation on overall clustering.
- The method's viability is demonstrated using default mode network analysis in individuals with schizophrenia (SZ).
Key Points:
- The novel fuzzy clustering approach more accurately identifies the importance of dFNC features to overall clustering compared to existing methods.
- Significant differences in default mode network state dynamics were identified between individuals with SZ and healthy controls.
- Interactions involving the posterior cingulate cortex (PCC) were found to be important across both the novel and existing approaches.
Conclusions:
- The developed explainable clustering approach offers a more comprehensive understanding of dFNC feature importance in rs-fMRI analysis.
- This method has the potential to advance research in neuropsychiatric disorders and other clustering applications.
- The findings highlight the utility of fuzzy clustering for analyzing complex brain network dynamics.
